Files
test/intern/cycles/kernel/kernels/opencl/kernel.cl
Jeroen Bakker 15edda3a8e T61463: Separate Baking kernels
Cycles OpenCL: Split baking kernels in own program

Fix T61463. Before this patch baking was part of the base kernels. There
are 3 baking kernels that and all 3 uses shader evaluation. Only for one
of these kernels the functionality was wrapped in the __NO_BAKING__
compile directive.

When you start baking this leads to long compile times. By separating
in individual programs will reduce the compile times.

Also wrapped all baking kernels with __NO_BAKING__ to reduce the
compilation times.

Impact on compilation time

    job   |   scene_name    | previous |  new  | percentage
  --------+-----------------+----------+-------+------------
   T61463 | empty           |    10.63 |  7.27 |         32%
   T61463 | bmw             |    17.91 | 14.24 |         20%
   T61463 | fishycat        |    19.57 | 15.08 |         23%
   T61463 | barbershop      |    54.10 | 48.18 |         11%
   T61463 | classroom       |    17.55 | 14.42 |         18%
   T61463 | koro            |    18.92 | 17.15 |          9%
   T61463 | pavillion       |    17.43 | 14.23 |         18%
   T61463 | splash279       |    16.48 | 15.33 |          7%
   T61463 | volume_emission |    36.22 | 34.19 |          6%

Impact on render time

    job   |   scene_name    | previous |   new   | percentage
  --------+-----------------+----------+---------+------------
   T61463 | empty           |    21.06 |   20.54 |          2%
   T61463 | bmw             |   198.44 |  189.59 |          4%
   T61463 | fishycat        |   394.20 |  388.50 |          1%
   T61463 | barbershop      |  1188.16 | 1185.49 |          0%
   T61463 | classroom       |   341.08 |  339.27 |          1%
   T61463 | koro            |   472.43 |  360.70 |         24%
   T61463 | pavillion       |   905.77 |  902.14 |          0%
   T61463 | splash279       |    55.26 |   54.92 |          1%
   T61463 | volume_emission |    62.59 |   39.09 |         38%

I don't have a grounded explanation why koro and volume_emission is this much
faster; I have done several tests though...

Maniphest Tasks: T61463

Differential Revision: https://developer.blender.org/D4376
2019-02-19 16:33:50 +01:00

149 lines
4.1 KiB
Common Lisp

/*
* Copyright 2011-2013 Blender Foundation
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
/* OpenCL kernel entry points - unfinished */
#include "kernel/kernel_compat_opencl.h"
#include "kernel/kernel_math.h"
#include "kernel/kernel_types.h"
#include "kernel/kernel_globals.h"
#include "kernel/kernel_color.h"
#include "kernel/kernels/opencl/kernel_opencl_image.h"
#include "kernel/kernel_film.h"
#if defined(__COMPILE_ONLY_MEGAKERNEL__) || !defined(__NO_BAKING__)
# include "kernel/kernel_path.h"
# include "kernel/kernel_path_branched.h"
#else /* __COMPILE_ONLY_MEGAKERNEL__ */
/* Include only actually used headers for the case
* when path tracing kernels are not needed.
*/
# include "kernel/kernel_random.h"
# include "kernel/kernel_differential.h"
# include "kernel/kernel_montecarlo.h"
# include "kernel/kernel_projection.h"
# include "kernel/geom/geom.h"
# include "kernel/bvh/bvh.h"
# include "kernel/kernel_accumulate.h"
# include "kernel/kernel_camera.h"
# include "kernel/kernel_shader.h"
#endif /* defined(__COMPILE_ONLY_MEGAKERNEL__) || !defined(__NO_BAKING__) */
#include "kernel/kernel_bake.h"
#ifdef __COMPILE_ONLY_MEGAKERNEL__
__kernel void kernel_ocl_path_trace(
ccl_constant KernelData *data,
ccl_global float *buffer,
KERNEL_BUFFER_PARAMS,
int sample,
int sx, int sy, int sw, int sh, int offset, int stride)
{
KernelGlobals kglobals, *kg = &kglobals;
kg->data = data;
kernel_set_buffer_pointers(kg, KERNEL_BUFFER_ARGS);
kernel_set_buffer_info(kg);
int x = sx + ccl_global_id(0);
int y = sy + ccl_global_id(1);
bool thread_is_active = x < sx + sw && y < sy + sh;
if(thread_is_active) {
kernel_path_trace(kg, buffer, sample, x, y, offset, stride);
}
if(kernel_data.film.cryptomatte_passes) {
/* Make sure no thread is writing to the buffers. */
ccl_barrier(CCL_LOCAL_MEM_FENCE);
if(thread_is_active) {
kernel_cryptomatte_post(kg, buffer, sample, x, y, offset, stride);
}
}
}
#else /* __COMPILE_ONLY_MEGAKERNEL__ */
__kernel void kernel_ocl_convert_to_byte(
ccl_constant KernelData *data,
ccl_global uchar4 *rgba,
ccl_global float *buffer,
KERNEL_BUFFER_PARAMS,
float sample_scale,
int sx, int sy, int sw, int sh, int offset, int stride)
{
KernelGlobals kglobals, *kg = &kglobals;
kg->data = data;
kernel_set_buffer_pointers(kg, KERNEL_BUFFER_ARGS);
kernel_set_buffer_info(kg);
int x = sx + ccl_global_id(0);
int y = sy + ccl_global_id(1);
if(x < sx + sw && y < sy + sh)
kernel_film_convert_to_byte(kg, rgba, buffer, sample_scale, x, y, offset, stride);
}
__kernel void kernel_ocl_convert_to_half_float(
ccl_constant KernelData *data,
ccl_global uchar4 *rgba,
ccl_global float *buffer,
KERNEL_BUFFER_PARAMS,
float sample_scale,
int sx, int sy, int sw, int sh, int offset, int stride)
{
KernelGlobals kglobals, *kg = &kglobals;
kg->data = data;
kernel_set_buffer_pointers(kg, KERNEL_BUFFER_ARGS);
kernel_set_buffer_info(kg);
int x = sx + ccl_global_id(0);
int y = sy + ccl_global_id(1);
if(x < sx + sw && y < sy + sh)
kernel_film_convert_to_half_float(kg, rgba, buffer, sample_scale, x, y, offset, stride);
}
__kernel void kernel_ocl_zero_buffer(ccl_global float4 *buffer, uint64_t size, uint64_t offset)
{
size_t i = ccl_global_id(0) + ccl_global_id(1) * ccl_global_size(0);
if(i < size / sizeof(float4)) {
buffer[i+offset/sizeof(float4)] = make_float4(0.0f, 0.0f, 0.0f, 0.0f);
}
else if(i == size / sizeof(float4)) {
ccl_global uchar *b = (ccl_global uchar*)&buffer[i+offset/sizeof(float4)];
for(i = 0; i < size % sizeof(float4); i++) {
*(b++) = 0;
}
}
}
#endif /* __COMPILE_ONLY_MEGAKERNEL__ */